WCD-04. Confronting the low summer CAPE behavior in GFSv16

Abstract
Convective Available Potential Energy (CAPE), when used in conjunction with other atmospheric properties, is an important forecasting variable to predict thunderstorms. One of the forecasting challenges in Global Forecast System version 16 (GFS v16) is the predicted too low summer CAPE values (Yang et al. 2020), which are smaller than forecast by the GFS v15.2 and indicated by the verifying Rapid Refresh (RAP) analysis. We use a summer case on Jul 23, 2020, which is included in the Unified Forecast System Case Studies collection, to investigate this low summer CAPE issue. Results show that the observed Mixed Layer CAPE at the Atmospheric Radiation Measurement central facility (C1) at the Southern Great Plains (SGP) at 00z Jul 24 exceeded 3,000 J/kg while the CAPE values from GFS v15.2 and GFS v16 simulations are 1730 and 1249 J/kg, respectively. We examine the GFS v16 model performance in simulating the meteorological fields, cloud, and surface energy budget at SGP C1 site. The tendencies of temperature, wind, and humidity created by the GFSv 15.2 and GFS v16 physics suites are examined by leveraging capabilities of the Common Community Physics Package (CCPP). We also investigate the forecast sensitivities to the planetary boundary layer parameterizations and surface fluxes specifications using the CCPP Single Column Model (SCM). Preliminary investigation shows that the low CAPE values in GFS are due to the drier air near the surface. Specifically, GFS v16 generates a lower 2-m dew point associated with a drier soil moisture than observation. The temperature vertical profiles are captured reasonably well in GFS. The simulated humidity profiles deviate from the sounding profiles measured at SGP C1 with larger dew points at higher altitudes and lower dew points at lower altitudes. GFSv16 underestimates the surface net radiation with excessive upper level clouds leading to less downward shortwave radiation in this case. In this presentation we will show forecast verification against analyses and observations, along with the sensitivity of SCM-simulated CAPE to the choice of physics schemes, to shed light on the reasons for the low CAPE bias in the GFS.